Separability

"As I've said, most of the autosomal tests will tell most people just what they see in the mirror every morning"

How so?

If there are very many Brazilian-type mixes, making up a very large proportion of the world population, the racial typologies become less useful. With even one such population, race becomes a much mushier category which is just the outcome of historical settlement and migration patterns. SE Asia, Hawaii, and E and N Africa strike me as possibly similiar cases. You'd still end up with relatively pure cases at the corners (Norway / Japan / Congo, I suppose) with some small exceptional peoples (Australian aborigines, !Kung....), but racial typologies don't seem like an especially powerful analytic tool.

Historically (into the XXc) language groups and political nations were mistaken for races, and this is why the word "race" got its bad name. Ideologically, race even became identified with a unit of land or territory, so various peoples ended up trying to figure out which was their real national homeland. Given the fact of extensive historical migration, this became insane, with more than one group often claiming the same homeland. It's really irrelevant -- if the Celts originally came from Central Europe about 2000 BC, e.g., why should the Irish care more than anyone else? It might be a mildly amusing conversational tidbit, but nothing more.

This connects to the point of the people trying to figure out whether they are really Hakka or really Sorbs, etc. It's not just that the tools we have aren't good enough yet. It seems pretty likely that, in most cases (especially farther back than about 200 years) there's really nothing much to know. There might be differences in distributions of various markers between populations of Poles, Swedes, Germans, English, and Irish, but if a marker is much more common among Poles than Irish, that doesn't mean that there's any meaningful connection between the Irishman with the marker and the Poles.

When people start talking about niche "races" like Hakka, it strikes me that the stupidity of XIXc racism is being revived: "One race, one language, one territory, once culture".

How so?

ancestrybyDNA tells people they are "indo-european" (white), etc. they already know this by induction from physical inspection of the geographical origin of ppl who look like them.

SE Asia, Hawaii, and E and N Africa strike me as possibly similiar cases. You'd still end up with relatively pure cases at the corners (Norway / Japan / Congo, I suppose) with some small exceptional peoples (Australian aborigines, !Kung....), but racial typologies don't seem like an especially powerful analytic tool.

one issue is that people tend to conceive many groups as "admixtures" of "pure races." to use an example, south asians are often conceived of as an admixture of caucasoids with other groups (australoid, east asian, etc.). the ancestryDNA tests tend to reinforce this perception. but the problem is that just because south asians lay in particular relation to west eurasians and east eurasians on a principal component chart when you look at the frequency of given alleles does not imply that south asians are simply an admixture in the biological sense of original pure populations. rather, they simply lay along a particular point on the clinal gradation as allele frequencies shift. one might be able to say the same of people in northeast africa, who might have always been genetically in between west africans and west eurasians...because they are geographically.

that does not mean that i think all these typologies are worthless, and the rate of change of allele frequencies is not constant, nor is the population density constant (so you have clumpiness in absolute numbers, with there being many more of group X and group Y than (X + Y)/2). you have to simply ask yourself what you are looking for. in the context of this post, i suspect people who are purchasing these tests want a more fine grained analysis than they will get. in some ways, this relates to my "mountainist vs. hillist" analogy, just because the boundary between the two is not well defined does not mean the terms are useless, but you have to be careful when people tell you that a mountain is at 1,000 feet and a hill as at 999.9999999.... feet as if that really carries any weight.

but if a marker is much more common among Poles than Irish, that doesn't mean that there's any meaningful connection between the Irishman with the marker and the Poles.

When people start talking about niche "races" like Hakka, it strikes me that the stupidity of XIXc racism is being revived: "One race, one language, one territory, once culture".

1) yes, i don't know if it is that meaningful in an everyday context to be 'genetically' polish or irish. on the other hand, i think it can tell us about historical questions when you test hypotheses. the saxon migration into britain is a case in point.

2) the problems in the second case are i suspect cognitive. human beings are naturally comfortable with the nature of the variation of our species in terms of generating rational categories because we are "meant" to live in groups of a few hundred. obviously racial typologies are not innate in the sense that they could never have really evolved in a situation where physical differences between tribes were minimal (which is why particular tatoos or markings or accents were important distinctions). but, we do have a fascination with our personal lineage and ancestry, and that is what these tests are tapping....

"ancestrybyDNA tells people they are "indo-european" (white), etc. they already know this by induction from physical inspection of the geographical origin of ppl who look like them."

The study by Rosenberg et al. (2002) shows differentiation within European and Asian populations. It's not like it isn't possible.

In reference to my earlier post:
http://img413.imageshack.us/img413/8686/ros20021ub.jpg

shogun, to make it short, i think the problem will be that people want to know how much french and spanish they are as individuals. so,

1) not only are the populations often rather similar
2) but separability on the populational level doesn't mean that individuals will be able to infer something particular significant

in other words, to give a positive case, an individual who is adopted, and looks partly native american and european american, would benefit from this sort of test as it might offer confirmatory evidence. on the other hand, some people want to know 'how much german' or 'how much french' they are, when there are usually records of where their ancestors come from via traditional genealogy. of course, that begs the question, 'what is a german.' on a population wide level there are differences between germans and the french...but the intersection is significant enough that i think that people will be misled if they find out that their ancestors were 'french speaking germans,' as the modal genetic profile of germans is simply taken to be archetypically german, blah, blah....

" i think the problem will be that people want to know how much french and spanish they are as individuals"

Yes, this is virtually impossible. To make a long story short: a South German from an isolated mountain village can easily be more related to a Frenchman who lives a few hundred kilometres away in another secluded village than to a North German because the two mountaineers (French and German) probably share a great deal of (pre-Indo-European) ancestry.